Re: BC break: ReflectionMethod::invoke() expects parameter1to be object, string given

On Tue, Aug 23, 2016 at 8:49 AM, Julien Pauli <jpauli@php.net> wrote: > On Tue, Aug 23, 2016 at 4:13 PM, Levi Morrison <levim@php.net> wrote: >> On Tue, Aug 23, 2016 at 7:56 AM, Alexander Lisachenko >> <lisachenko.it@gmail.com> wrote: >>> >>> 2016-08-23 16:40 GMT+03:00 Julien Pauli <jpauli@php.net>: >>>> >>>> My patch allows that, but I can't find a use case to it. >>> >>> >>> >>> My use case for that was decorating of static methods with additional >>> behaviour, so it's possible to cache the result of static methods, etc. >>> Unfortunately, reflection API doesn't provide me such an ability, so I >>> decided to switch to the combination of special closure with >>> forward_static_call_array() and binding it to the desired scope. >>> >>> >>> 2016-08-23 16:40 GMT+03:00 Julien Pauli <jpauli@php.net>: >>>> >>>> Try it. >>> >>> >>> Yes, sure, I will check. However as I say earlier, now I use closure >>> extraction and scope binding as more natural way to invoke such methods. So >>> missing scope argument for static methods and ReflectionMethod->invoke() is >>> a nice thing to have, but not a critical one. >> >> Seems to me you should be calling getClosure() and using >> Closure::bindTo before invoking it. ReflectionMethod::invoke is really >> just a short-cut for a common case; I don't think it's reasonable to >> have it mirror the Closure API just to save one method call to get a >> Closure. > > This cannot work. > There is a special use case forbidding that. > "Cannot rebind scope of closure created by > ReflectionFunctionAbstract::getClosure()" > > > Julien I know this is a stability concern for internal classes but is there anything really preventing this from user-land classes?